Nonparametric estimation of a mixture of probability densities of signals (interferences)

E. A. Ibatoulline

Institute of Physics, Kazan (Volga region) Federal University

Abstract: In this article, we suggest using the parallel clustering procedure for the nonparametric estimation of a mixture of probability densities of signals (interferences). This procedure makes it possible to estimate with an adequate accuracy the parameters of the marginal mixture distribution, as well as the number of components (classes) in a mixture. The efficiency of the suggested method is confirmed by the numerical results.

Keywords: nonparametric estimation, mixture of probability densities, parallel clustering procedure.

UDC: 621.391.26

Received: 02.04.2010